Critical infrastructures like the power grid are emerging as collection of existing separated systems of different nature which are interconnected together. Their criticality becomes more and more evident as the damage and the risks deriving from wrong behaviors (both accidental and intentionally caused) are increasing. It is becoming evident that existing (legacy) subsystem must be interconnected together following some disciplined and controlled way. This is one of the challenges taken by the European Project CRUTIAL, where an infrastructure architecture seen as a WAN of LANs is being proposed, where LANs confine existing sub-systems, protected by special interconnection and filtering devices (CIS - CRUTIAL Information Switches). Previous work led to the definition of the CIS internal and interconnection architecture, so that a set of CIS can collectively ensure that the computers controlling the physical process correctly exchange information despite accidents and malicious attacks. CIS resilience is achieved thanks to replication for intrusion tolerance and replica recovery for self-healing. This chapter analyzes the redundant architecture of the CIS, with a set of objectives: identifying the relevant parameters of the architecture; evaluating how effective is the trade-off between proactive and reactive recoveries; and finding the best parameter setup. Two measures of interest were identified, a model of the recovery strategy was constructed and the quantitative behavior of the recovery strategy was analyzed. The impact of the detection coverage, of the intrusions and of the number of CIS replicas was analyzed and discussed. The directions for refining and improving the recovery strategy were proposed. © Springer-Verlag Berlin Heidelberg 2008.
|Original language||English (US)|
|Title of host publication||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|Number of pages||23|
|State||Published - Sep 22 2008|
ASJC Scopus subject areas
- Theoretical Computer Science
- Computer Science(all)